. 2024 Feb 14;25:69. doi: 10.1186/s12859-024-05679-9

Table 2.

Simulation with variable–variable connections: randomly select combinations of hyper-parameters to search over

Method	Error (%)	TPR-1	TPR-2	FPR-1	FPR-2	F-1	F-2
Scale-free
Setting 1
$(p_{1} = 500, p_{2} = 500, n_{1} = 200, n_{2} = 150)$
iDeepViewLearn	6.84 (1.96)	95.71	96.19	0.19	0.17	95.71	96.19
iDeepViewLearn-Laplacian	7.10 (1.65)	97.86	98.57	0.09	0.06	97.86	98.57
Sparse CCA + SVM	21.80 (10.14)	100.00	100.00	14.12	15.13	42.97	43.26
Fused CCA + SVM	41.33 (7.52)	19.29	23.10	17.90	33.29	6.56	4.08
Deep CCA + TS + SVM	40.43 (1.61)	5.00	3.33	4.16	4.24	5.00	3.33
MOMA	45.55 (1.84)	17.14	25.95	3.63	3.25	17.14	25.95
MOMA + SVM	36.76 (5.41)	17.14	25.95	3.63	3.25	17.14	25.95
Random Forest on stacked data	11.99 (2.22)	52.38	85.23	2.09	0.65	52.38	85.23
SVM on stacked data	35.46 (1.23)	–	–	–	–	–	–
Setting 2
$(p_{1} = 500, p_{2} = 500, n_{1} = 6000, n_{2} = 4500)$
iDeepViewLearn	2.77 (0.62)	99.76	100.00	0.01	0.00	99.76	100.00
iDeepViewLearn-Laplacian	2.71 (0.14)	100.00	100.00	0.00	0.00	100.00	100.00
Sparse CCA + SVM	9.25 (2.39)	100.00	100.00	9.35	9.22	56.74	52.59
Fused CCA + SVM	33.24 (4.83)	99.75	100.00	13.69	48.83	48.74	19.90
Deep CCA + TS + SVM	2.44 (0.31)	17.62	17.38	3.61	3.62	17.62	17.38
MOMA	41.88 (4.11)	63.81	72.62	1.59	1.20	63.81	72.62
MOMA + SVM	3.56 (4.37)	63.81	72.62	1.59	1.20	63.81	72.62
Random Forest on stacked data	1.86 (0.10)	100.00	100.00	0.00	0.00	100.00	100.00
SVM on stacked data	27.61 (0.23)	–	–	–	–	–	–
Lattice
Setting 1
$(p_{1} = 500, p_{2} = 500, n_{1} = 200, n_{2} = 150)$
iDeepViewLearn	4.90 (1.76)	100.00	98.88	0.00	0.12	100.00	98.88
iDeepViewLearn-Laplacian	3.90 (0.82)	99.80	99.59	0.02	0.04	99.80	99.59
Sparse CCA + SVM	16.03 (0.86)	100.00	100.00	1.29	1.45	94.96	94.77
Fused CCA + SVM	38.26 (11.58)	22.04	24.69	24.59	30.99	11.13	9.30
Deep CCA + TS + SVM	36.53 (2.05)	10.61	10.71	9.71	9.70	10.61	10.71
MOMA	44.76 (2.12)	23.98	26.53	8.26	7.98	23.98	26.53
MOMA + SVM	32.20 (4.27)	23.98	26.53	8.26	7.98	23.98	26.53
Random Forest on stacked data	3.41 (0.71)	66.84	92.86	3.60	0.78	66.84	92.86
SVM on stacked data	28.51 (0.56)	–	–	–	–	–	–
Setting 2
$(p_{1} = 500, p_{2} = 500, n_{1} = 6000, n_{2} = 4500)$
iDeepViewLearn	1.64 (0.17)	100.00	100.00	0.00	0.00	100.00	100.00
iDeepViewLearn-Laplacian	1.56 (0.12)	100.00	100.00	0.00	0.00	100.00	100.00
Sparse CCA + SVM	7.14 (2.92)	100.00	100.00	1.14	1.72	95.52	93.52
Fused CCA + SVM	5.26 (2.27)	100.00	100.00	3.44	6.84	88.89	78.95
Deep CCA + TS + SVM	0.98 (0.19)	39.49	32.04	6.57	7.38	39.49	32.04
MOMA	21.22 (13.01)	73.98	79.08	2.83	2.27	73.98	79.08
MOMA + SVM	1.27 (1.53)	73.98	79.08	2.83	2.27	73.98	79.08
Random Forest on stacked data	1.02 (0.07)	100.00	100.00	0.00	0.00	100.00	100.00
SVM on stacked data	8.57 (0.24)	–	–	–	–	–	–
Cluster
Setting 1
$(p_{1} = 500, p_{2} = 500, n_{1} = 200, n_{2} = 150)$
iDeepViewLearn	22.50 (1.73)	96.21	100.00	0.27	0.00	96.21	100.00
iDeepViewLearn-Laplacian	22.40 (2.14)	95.15	100.00	0.34	0.00	95.15	100.00
Sparse CCA + SVM	16.70 (1.22)	100.00	100.00	4.24	3.91	77.50	78.54
Fused CCA + SVM	43.27 (1.65)	16.97	16.06	18.96	22.76	5.52	5.56
Deep CCA + TS + SVM	37.96 (1.81)	7.12	6.97	6.56	6.57	7.12	6.97
MOMA	45.28 (2.02)	21.52	23.18	5.55	5.43	21.52	23.18
MOMA + SVM	36.61 (3.77)	21.52	23.18	5.55	5.43	21.52	23.18
Random Forest on stacked data	29.23 (1.19)	27.42	65.76	5.13	2.42	27.42	65.76
SVM on stacked data	31.60 (1.02)	–	–	–	–	–	–
Setting 2
$(p_{1} = 500, p_{2} = 500, n_{1} = 6000, n_{2} = 4500)$
iDeepViewLearn	15.78 (0.65)	100.00	99.39	0.00	0.04	100.00	99.39
iDeepViewLearn-Laplacian	15.70 (0.35)	96.21	100.00	0.27	0.00	96.21	100.00
Sparse CCA + SVM	14.59 (0.53)	100.00	100.00	12.07	7.54	57.31	66.46
Fused CCA + SVM	29.17 (9.55)	72.73	92.42	26.12	30.40	31.17	41.73
Deep CCA + TS + SVM	28.48 (1.52)	10.45	8.64	6.33	6.46	10.45	8.64
MOMA	39.22 (4.95)	73.18	91.82	1.90	0.58	73.18	91.82
MOMA + SVM	12.77 (0.72)	73.18	91.82	1.90	0.58	73.18	91.82
Random Forest on stacked data	13.83 (0.21)	100.00	100.00	0.00	0.00	100.00	100.00
SVM on stacked data	29.68 (0.22)	–	–	–	–	–	–

TPR-1; true positive rate for $X^{(1)}$ . Similar for TPR-2. FPR; false positive rate for $X^{(2)}$ . Similar for FPR-2; F-1 is the F measure for $X^{(1)}$ . Similar for F-2. The highest F-1/2 is in red. (The mean error of two views is reported for MOMA; MOMA + SVM means combining the feature selection part of MOMA and SVM)